A telephone poll was conducted by Fleming and Associates of Cumberland (sponsored by WPRI-12 News and Roger Williams University) of 402 RI likely voters, 254 of whom were in the Congressional second district, from Thu, Sep 29, to Sun, Oct 2.
The most notable result shows Congressional second district Republican nominee Allan Fung leading Democratic nominee Seth Magaziner, 46% – 40%, with independent candidate William Gilbert at 4%, not sure at 9%, and refused to answer at 1%. Due to the small number polled, the sampling margin of error is a relatively large ±6.2 percentage points, meaning that the actual numbers in the population under study could, in fact, be exactly reversed.
There are other elements of statistical bias in the poll, particularly the high percentage of cellular (90%) as opposed to landline (10%) respondents, which tends to oversample younger voters and undersample older voters. Age-bias is important because the same poll shows Magaziner leading Fung 49% – 37% among voters 18 – 39 but Fung leading Magaziner 49% – 39% among voters 40 and older. Counter-intuitively, this means the sampling margin of error for undersampled subpopulations, once the raw results are normalized and weighted to compensate, could be easily an order of magnitude greater than that of the overall sample, such that only a handful of outlier responses would greatly distort the reported result. In other words, Magaziner is doing much better among the subpopulations that the poll captures more accurately. The poll does not publish sufficient raw data needed to estimate the boundaries of this problem.
As Motif noted a few days ago just before these poll results were released (“News Analysis — Elections 2022: Few contested races remain after primaries,” by Michael Bilow, Oct 5), although “Fung is an unusually strong Republican candidate by RI standards and has a credible chance of winning the House seat,” based on historical data “political forecasting news service Five Thirty Eight considers the race ‘likely Democratic’ with an 83% probability of Magaziner winning…”
The same poll reports the race for governor with Democratic nominee Daniel McKee leading Republican nominee Ashley Kalus by a double-digit margin, 45% – 32%, with not sure at 15%, refused to answer at 1%, and the remainder scattered among independent Paul Rianna (3%), independent Zachary Hurwitz (2%), and libertarian Elijah Gizzarelli (2%). For the statewide race, the sampling margin of error is a tighter ±4.9 percentage points. This is consistent with the prior observation in Motif that “the conventional wisdom is that Kalus has the chance of a snowball in hell of upsetting McKee.”
The poll attempted “favorability” ratings for McKee and Kalus: reporting McKee at 13% very favorable, 32% somewhat favorable, 14% somewhat unfavorable, 22% very unfavorable, and 19% not sure; reporting Kalus at 16% very favorable, 17% somewhat favorable, 13% somewhat unfavorable, 19% very unfavorable, and 35% not sure. The usual way of summarizing this is to view McKee as net +9 (45% favorable – 36% unfavorable) and Kalus as net +1 (33% favorable – 32% unfavorable). The large “not sure” blocs are more of an opportunity for Kalus to define herself to the one-third of the electorate who pretty much has no idea who she is, while a concern for McKee who as the sitting governor should be worried that one-fifth of the electorate does not know enough about him to form an opinion.
Asking about the direction the state has been moving, 34% said right, 45% said wrong, 9% said neither right nor wrong, and 12% said not sure. A question like this is minimally useful, essentially measuring satisfaction or dissatisfaction with the current political leadership.
Asked to choose the single most important issue from a list, 42% said cost of living, 14% said abortion, 11% said education, 10% said health care, 9% said taxes, 7% said public safety, 6% said other, and 2% answered not sure.
Public money for construction of a soccer stadium in Pawtucket was overwhelmingly unpopular, 56% – 31% opposed with 12% not sure.
